EFuse
An e-fuse (electronic fuse) sits in series with a power rail and monitors the current. When current exceeds the programmed limit, it turns off the rail in microseconds to protect downstream circuitry and the supply. Unlike a physical fuse, it can be re-enabled in software.
An internal MOSFET and a precision current-sense amplifier form a fast control loop. A user-resistor sets the overcurrent trip point. When current exceeds that point, the MOSFET shuts off; a fault flag goes low on an I²C or GPIO line. Some devices also slew the inrush current to prevent supply collapse on hot-plug.
In plain terms
A circuit breaker that resets with software instead of a push button — it monitors current in real time and opens in microseconds if the load misbehaves, then a microcontroller can retry it once the fault is cleared.
Why designers use it
- Protect backplane slots from a misbehaving plug-in card drawing excess current.
- Limit inrush from large capacitors on a hot-plugged module.
- Allow firmware to remotely power-cycle a misbehaving subsystem.
- Replace mechanical fuses in applications where field-replacement is impractical.
Best for
- Server backplanes
- USB ports
- Hot-plug modules
Key specifications
- Vin: 0.8 V – 60 V
- Current limit: 0.5 A – 20 A (Programmable)
- Rds(on): 10 – 100 mΩ
- Response time: 1 µs – 100 ms
- Auto-retry: Latch-off or auto-restart options
When not to use it
- When the downstream fault current could rise faster than the e-fuse can respond (< 1 µs faults may need a physical fuse upstream).
- Where the MOSFET's on-state voltage drop is unacceptable at very high currents.
Common mistakes
- Setting the current limit too close to the normal operating current, causing nuisance trips on inrush.
- Ignoring the thermal rise in the internal MOSFET during current limiting — check the SOA curve at the expected fault current and duration.
Where you will find it
- A USB hub uses an e-fuse on each downstream port: when a faulty device short-circuits the 5 V bus, the e-fuse shuts that port off in 2 µs, protecting the hub IC and keeping the other ports live — something a shared polyfuse cannot do.
- A server's hot-swap card slot uses an e-fuse to control inrush as a new PCIe card is inserted: the IC ramps the current slowly over 10 ms, preventing the 100 µF bulk capacitors on the new card from dragging the 12 V rail down and resetting other servers sharing the bus.
- An industrial robot's tool-changer power outlet uses an e-fuse so the robot controller can power-cycle a jammed end-effector by toggling a GPIO — no operator access to the joint is required, and the e-fuse's latch-off prevents continuous fault current if the new tool is still shorted.
A short history
The electronic fuse (eFuse) IC integrates a power MOSFET, current-sense circuitry, and control logic to emulate — and improve upon — the function of a traditional fuse. Texas Instruments introduced one of the earliest dedicated eFuse ICs in the late 1990s, targeting telecom and server power systems where replacing a blown fuse required a service call. Unlike a passive fuse that permanently opens on overcurrent, an eFuse limits current, protects against voltage transients, and automatically retries or holds off until the fault is cleared, then resets without human intervention. Programmable current thresholds, slew-rate control for inrush limiting, and diagnostic reporting via I²C or SMBus make the modern eFuse a critical component in hot-plug server cards, USB ports, and industrial field instruments.
Good to know
- An e-fuse replaces a glass-tube fuse with an integrated MOSFET + current-sense + comparator — it trips in microseconds and resets at the press of a button.
- USB-C ports use e-fuses with programmable current limits (1.5 A / 3 A / 5 A) so the same hardware can negotiate different power-delivery profiles.
- Modern hot-swap e-fuses also limit inrush current (dV/dt control), preventing the 'bulk capacitor charge' from latching the upstream rail.